Opus Research has long tracked voice as a natural user interface to trusted, convenient interactions between businesses and their customers. At the same time, a predictable consequence of “voice-first” adoption has been increased malicious attempts to compromise speech-enabled channels by fraudsters.

To combat these attempts and improve security defenses, many technology providers are blending the ever-increasing compute power of Big Data with Deep Neural Networks (DNNs) to deliver breakthrough performances in Voice Biometrics and create critical solutions for intelligent authentication and fraud prevention.

Last week, Czech-based Phonexia announced the release of the fourth-generation voice biometrics product, Deep Embeddings for Speaker Identification. Utilizing accelerated processing power with DNN, the company boasts robust speaker modules, improved equal error rates (EER) and less time needed for speaker identification.

In the announcement, Phonexia says these advancements will help traditional clients adopt voice biometric technologies for fraud prevention but expects to see adoption in “new segments such as 4.0 devices, automotive, smart wearables, IoT devices, and devices with no permanent connection to the Internet,” said Petr Schwarz, CTO with Phonexia.

The company is focused on combining speaker identification technologies with data mining, speech transcription, and language identification in the Phonexia Speech Platform for a number of use cases in intelligent authentication, fraud detection, criminal investigation and public safety.

Categories: Conversational Intelligence, Intelligent Authentication, Articles